49Aglobalatlasofthedominantbacteriafoundi

BIOGEOGRAPHYAglobalatlasofthedominantbacteriafoundinsoilManuelDelgado-Baquerizo,1,2*AngelaM.Oliverio,1,3TessE.Brewer,1,4AlbertoBenavent-González,5DavidJ.Eldridge,6RichardD.Bardgett,7FernandoT.Maestre,2BrajeshK.Singh,8,9NoahFierer1,3*Theimmensediversityofsoilbacterialcommunitieshasstymiedeffortstocharacterizeindividualtaxaanddocumenttheirglobaldistributions.Weanalyzedsoilsfrom237locationsacrosssixcontinentsandfoundthatonly2%ofbacterialphylotypes(~500phylotypes)consistentlyaccountedforalmosthalfofthesoilbacterialcommunitiesworldwide.Despitetheoverwhelmingdiversityofbacterialcommunities,relativelyfewbacterialtaxaareabundantinsoilsglobally.Weclusteredthesedominanttaxaintoecologicalgroupstobuildthefirstglobalatlasofsoilbacterialtaxa.Ourstudynarrowsdowntheimmensenumberofbacterialtaxatoa“mostwanted”listthatwillbefruitfultargetsforgenomicandcultivation-basedeffortsaimedatimprovingourunderstandingofsoilmicrobesandtheircontributionstoecosystemfunctioning.Althoughsoilbacteriahavebeenstudiedformorethanacentury,mostofthediversityofsoilbacteriaremainsundescribed.Thisisunsurprisinggiventhatsoilbacteriarankamongthemostabundantanddi-versegroupoforganismsonEarth(1–4),chal-lengingourcapacitytounderstandtheirspecificcontributionstoecosystemprocesses,includingnutrientandcarboncycling,plantproduction,andgreenhousegasemissions(1–3).Putsimply,characterizingtheecologicalattributes(environ-mentalpreferencesandfunctionaltraits)ofthethousandsofbacterialtaxafoundinsoilisun-feasible.Mostsoilbacteriadonotmatchthosefoundinpreexisting16SribosomalRNA(rRNA)genedatabases(5),wehavegenomicinforma-tionforrelativelyfewofthem(5–7),andthema-jorityofsoilbacteriahavenotbeensuccessfullycultivatedinvitro(6,7).Forthesereasons,welackapredictiveunderstandingoftheecolog-icalattributesofmostsoilindividualbacterialtaxa,withtheirenvironmentalpreferences,traits,andmetaboliccapabilitiesremaininglargelyunknown.Previousworkhasshownthatonlyasmallfractionofsoilbacteriaistypicallysharedbetweenanypairofuniquesoilsamples(4,8,9).However,wealsoknowthat,aswithmost“macrobial”com-munities(10),notallbacterialtaxaareequallyabundantinsoil.Thereareoftensubsetsofsoilbacterialtaxathatarefarmoreabundantthanothers.Forexample,thegenusBradyrhizobiumhasbeenfoundtobedominantinforestsoilsfromNorthAmerica(11).Similarly,alineagewithintheclassSpartobacteriawasfoundtobehighlyabundantinundisturbedgrasslandsoils(12).Perhapsmoreimportant,manyindividualtaxathatarehighlyabundantinindividualsoilsam-plesmayalsobeabundantacrossdistinctsoilsamples,evenwhenthosesoilsamplesarefromsiteslocatedfarapart(e.g.,CandidatusUdaeobactercopiosus)(13).Therefore,acriticalandlogicalnextsteptoadvanceourunderstandingofsoilbac-terialcommunitiesistoidentifythedominantbacterialphylotypesthatareabundantandubiq-uitousacrosssoils,anddeterminetheirecologi-calattributes.Fromthelargebodyofliteratureusingmarkergenesequencingtocharacterizesoilbacterialcommunities,weknowwhichmajorphylatendtobemoreabundantinsoil(14)andwehaveagrowingunderstandingofhowvariousfactors,includingsoilproperties(e.g.,pH)(15),climate(9,16),vegetationtype(17),andnutrientavail-ability(18),structurethecompositionofsoilbac-terialcommunitiesworldwide.Whatiscurrentlymissingisadetailedecologicalunderstandingofcommonsoilbacterialspecies,whichwerefertoasphylotypes(asbacterialspeciesdefinitionscanbeproblematic)(19).Understandingtheeco-logicalattributesofdominantphylotypeswillincreaseourabilitytosuccessfullycultivatetheminvitroandallowustobuildamorepredictiveunderstandingofhowsoilbacterialcommu-nitiesvaryacrossspace,time,andinresponsetoanthropogenicchanges.Forexample,ifwecouldidentifythosedominantphylotypeswithstrongpreferencesforagivensetofenviron-mentalconditions(e.g.,loworhighpH),wecouldthenusethisinformationtopredicttheirdistri-butionsandenrichforthesedominantphylotypesinvitro.Ultimately,abetterunderstandingofdominantsoilbacterialtaxawillimproveourabilitytoactivelymanagesoilbacterialcommu-nitiestopromotetheirfunctionalcapabilities.Hereweconductedaglobalanalysisofthebacterialcommunitiesfoundinsurfacesoilsfrom237locationsacrosssixcontinentsand18coun-tries(fig.S1)to(i)identifythemostdominant(i.e.,mostabundantandubiquitous)soilbacte-rialphylotypesworldwide;(ii)determinewhichofthesedominantphylotypestendtoco-occurandsharesimilarenvironmentalpreferences;(iii)maptheabundancesoftheseecologicalclus-tersofdominantsoilbacteriaacrosstheglobe;and(iv)assessthegenomicattributesthatdif-ferentiatephylotypeswithdistinctenvironmen-talpreferences.Thesoilsincludedinthisstudywereselectedtospanawiderangeofvegetationtypes,edaphiccharacteristics,andbioclimaticregions(arid,temperate,tropical,continentalandpolar)(20).Wefirstidentifiedthemostdominantbacte-rialphylotypesby16SrRNAgeneampliconse-quencing(20).Dominantphylotypes(taxathatshare≥97%sequencesimilarityacrosstheampli-fied16SrRNAgeneregion)includethosethatarehighlyabundant(top10%mostcommonphylotypessortedbytheirpercentageof16SrRNAreads)(21)andubiquitous(foundinmorethanhalfofthe237soilsamplesevaluated)(20).Notsurprisingly,ourglobaldat